ML Research Engineer

Unitary’s purpose is to make the internet safer to improve lives and create a fairer world. Our mission is to develop video understanding AI and deploy it at scale to solve real-world problems in content moderation and beyond. We believe in an online world free from harm, where we can trust AI to make safe and fair decisions.

Unitary is a remote-first team of 25+ passionately curious individuals who want to improve our online world by understanding its content using AI. We are on the cusp of something very exciting, and are about to enter a period of explosive growth. We’re looking for people who enjoy solving serious challenges with determination and playfulness, just like us! You will be working with a group of kind and collaborative people in a tight-knit, fast-moving and international team. We are looking for someone who is as excited about Unitary’s mission as we are, who wants to have a large impact at an early-stage startup, and be a key part of defining Unitary’s future as one of our early employees. We need versatile people who are happy to get stuck into whatever needs doing, and are ready to learn and grow with the company.

We are now looking for more ML Research Engineers to join our growing research team, building multimodal foundation models for general content understanding. We believe that, due to the complexity and nuance of user-generated online content, completely automated moderation requires large-scale general multimodal models with human-level capabilities for reasoning and understanding. Our goal is to build multimodal foundation models that show a general understanding of text, images, video and audio. If this sounds interesting and relevant to you and your experience, please reach out!

As part of this role, you will:

• Develop, implement and train foundation models for general multimodal content understanding
• Evaluate model capabilities on a variety of tasks and datasets
• Ensure our training datasets have high-quality and diverse content
• Share knowledge through presentations, blog posts or publishing in top-tier ML conferences or journals
• Collaborate with product teams to develop solutions

Requirements

We would love to hear from you if you have:

• Experience training foundation models on large datasets
• Familiarity with large scale distributed training strategies
• Proficiency in Python and ML frameworks such as Pytorch or Jax
• Strong software engineering practices
• Passion for building systems from scratch
• Awareness of responsible AI development including privacy and fairness issues / Experience with benchmarking for fairness and distribution shifts

It would be even better, but not essential, if you have:

• Experience using SLURM
• Experience training multimodal foundation models for video
• Demonstrated significant contributions to relevant projects e.g. published papers at ML venues such as CVPR, ACL, NeurIPS, ICML, ICCV, EMNLP etc. or contributions to open source projects

Benefits

Unitary is a remote-first company that believes happy people are a must for company success. We don't compromise when it comes to creating a great culture built on genuine interest for each other's well being and shared positive experiences.

We are fiercely passionate about making the internet a safer place, and deeply motivated to become a force for good. We believe that as individuals we are awesome, but as a team, we are unbeatable.

We offer progressive and market leading benefits, which we hope will enable everyone to create well balanced and exciting lives both on and off work. To name a few of them:

• Flexible hours and location
• Flexible public holidays (choose to take public holidays on the days they fall, or take them at a different time of the year when it fits your values, beliefs, and calendar better)
• Share options

• Occupational pension
• Paid parental leave (3 months of fully paid for everyone regardless of gender, location or family situation)
• Paid sick leave (2 weeks fully paid)
• Brain-Love Budget (£1,000 per person per year towards professional development and growth)
• Feel-Good Budget (£500 per person per year towards anything health and wellness related)

Location: Fully remote / London